Joint Optimization of Run-Length Coding, Huffman Coding and Quantization Table with Complete Baseline JPEG Compatibility

Abstract

JPEG optimization strives to maximize the best rate distortion performance while remaining faithful to the JPEG syntax. Given an image, if soft decision quantization (SDQ) is applied to its DCT coefficients, then Huffman table, quantization step sizes and SDQ coefficients are three free parameters over which a JPEG encoder can optimize. In this paper, we first propose a novel algorithm to find the optimal SDQ coefficient indices in the form of run-size pairs among all possible candidates given that the other two parameters are fixed. Based on this algorithm, we then formulate an iterative algorithm to jointly optimize the run-length coding, Huffman coding and quantization step sizes. The proposed iterative algorithm achieves a compression performance better than any previously known JPEG compression results and even exceeds the quoted PSNR results of some state-of-the-art wavelet-based image coders like Shapiro's embedded zerotree wavelet algorithm at the common bit rates under comparison.